why does exec-shield load .so @ diff addrs run to run
From: Eric Taylor (et1_at_rocketship1.com)
Date: 04/28/04
- Next message: Grant Edwards: "Re: State of the GnuCash project, a call for help"
- Previous message: aw: "State of the GnuCash project, a call for help"
- Next in thread: Sean Caulfield: "Re: why does exec-shield load .so @ diff addrs run to run"
- Reply: Sean Caulfield: "Re: why does exec-shield load .so @ diff addrs run to run"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Wed, 28 Apr 2004 15:44:20 GMT
We have started trying out Fedora and have found that with exec-shield enabled (value=1) the load address of some .so libraries keep
moving around (slightly) from run to run. I was wondering why this should happen. Why isn't the algorithm deterministic? We have
been forced to turn it off, because it breaks our fast memory checkpoint/restart algorithm which depends on program libraries
loading into memory at the same place fro run to run.
I have seen a .so file load a few 10's of k bytes higher or lower, even leaving a small holes between adjacent .so loads.
I understand the ascii-armor trick, and a little bit of how exec-shield tracks high segment addresses, and I do understand the
desire to keep the stack from being executable, but I didn't see anything that would explain why it should work differently from run
to run. Is this supposed to happen?
- Next message: Grant Edwards: "Re: State of the GnuCash project, a call for help"
- Previous message: aw: "State of the GnuCash project, a call for help"
- Next in thread: Sean Caulfield: "Re: why does exec-shield load .so @ diff addrs run to run"
- Reply: Sean Caulfield: "Re: why does exec-shield load .so @ diff addrs run to run"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|